BOHR International Journal of Engineering

2022, Vol. 1, No. 1, pp. 28–32

DOI: 10.54646/bije.2022.06
www.bohrpub.com

System identification and characteristics of multiple input

and multiple output (MIMO) water mixing equipment
Kirubanandan Shanmugam *
Department of Biotechnology, Sree Sastha Institute of Engineering and Technology (Affiliated to Anna University, Chennai),
Chennai, India

*Correspondence:
Kirubanandan Shanmugam,
Kirubanandan.shanmugam@gmail.com

Received: 05 August 2022; Accepted: 25 August 2022; Published: 02 September 2022

Multiple input and multiple output (MIMO) water mixing equipment is a prototype used to study the process
control concepts. This equipment is used for manipulating multiple inputs and their effect on the outputs. The
input variables are hot water (HW) temperature, cold water (CW) temperature, and flow rate of HW and CW. The
output variables are the level of the water in the tank, the temperature of the water, and the flow of outlet water. The
changes in controlled variables with the effect of manipulated variables were investigated. The transfer function for
the given experimental setup was developed and the time constant for the given system was also developed. These
transfer function and analysis of MIMO can be useful for the analysis of various MIMO-based process equipments.
Keywords: controlled variables, manipulated variables, transfer function, time constant and MIMO prototype

Introduction done by observing the relationships between CW-to-level,

The multiple input and multiple output (MIMO) water
mixing-tank apparatus is an education prototype for
studying the fundamental concepts of process control. In this
system, there would be many control (output) variables that
are affected/controlled/manipulated by several manipulated
(input) variables used in a given process. The MIMO water
mixing-tank apparatus is illustrated in Figures 1, 2, 3, 4.
The controlled variables were two water supplies going into
the tank, namely, one hot water (HW) inlet and one cold
water (CW) inlet. These were controlled with the LabVIEW
program. The drain leaving the tank was left in the fully
open position throughout the experiment. The measured
variables (tank level and water temperature) were measured
with the LabVIEW program. The disturbance variables for
the system were the temperatures of the inlet streams [HW
temperature (Th) and CW temperature (Tc)]. Most industrial
control applications involve in MIMO. Modeling MIMO
processes is not different conceptually from modeling single-
input/single-output (SISO) processes (1).
The objective of this experiment was to identify key system
characteristics such as outlet flow, gain, delay, and time
constants for MIMO water mixing-tank apparatus. This was
HW-to-level, CW-to-temperature, and HW- to-temperature.
These relationships were examined with the LabVIEW
program. This program is a graphical programming platform
that is useful to scale from design to test and from small
to large systems. It is ideal for the measurement of control
systems (2). The experimental data were recorded in the
Excel format with Data Logger.
This experiment investigates some of the main key
characteristics such as the outflow, delay, gain, and the time
constant in the system. The experimental setup introduced
the concept of controlled systems. The setup contains a
large tank that has water in it as well as three manipulated
variables. The three manipulated variables are CW, HW, and
drain valves. The system characteristics were used to obtain a
relationship that relates the level of liquid in the tank and the
temperature of the system with step changes in the CW and
HW flows. The experiment contains several requirements
that were assigned to observe the variations of different
degrees of opening valves. The experiment was categorized
into 5 different degrees of obstruction of the flow of the
valves which demonstrates multiple observations that deal
with the height of liquid in the tank as well as the temperature
changing due to the abovementioned manipulated variables.

28
10.54646/bije.2022.06 29

FIGURE 1 | Schematic of the experimental setup.

FIGURE 3 | Hot and cold valves.

FIGURE 2 | The experimental setup.

In theory, the actuators in the experiment are the valves

(hot and cold) that are manipulated to obtain a stable output.
The stable output, whether it is the temperature or the
height/level of the liquid in the tank, is both plotted on
the screen. A step input is inserted through one of the
manipulated variables while maintaining the other valve at a
fixed value. This is done for the two manipulated variables to
produce an output response for each of the MV. The concept
FIGURE 4 | Drain valve.
of input delay or dead time that occurs in this experiment is
specifically concerned with the time required for the output
to respond to the changes in inlet variables. Clearly, it is the
dead time for a change in the level of liquid in the tank to
30
be observed on the screen. Furthermore, the parameters of
interest are the time constant and the gain that are obtained
in this experiment and evaluated later on. The time constant
represents the time required to reach 63% of the final value
or set point (3–5).
This article deals with the system identification and
analysis of MIMO apparatus to understand the basic concept
of process control and its implementation in the process
equipment and to evaluate the transfer function of the
MIMO apparatus. In addition to that, the interaction
between controlled and manipulated variables in the MIMO
apparatus is investigated.
Experimental setup and methods
Experimental procedure
Initially, the tank setup was at a steady state. This was obvious
from the leveling of the curve describing the height, h, and
the temperature, T, of the tank. The reading on the monitor
that correlates the degree of obstruction of the flow of the
valves is 0 closed, 1–1/5 opened, 2–2/5 opened, 3–3/5 opened,
4–4/5 opened, and 5 fully opened.
At t = 0, the drain valve was fully opened (Figure 5)
(reading 5 on the monitor), and the readings of cold valve
(CV) and hot valve (HV) were 3 and 2, accordingly. The
readings of h and T were recorded for this steady state.
Then, the CV’s reading was manually imputed to 2, while the
HV’s reading was maintained at 2. Then, a steady state was
achieved after about 10 min of waiting time. The steady-state
readings of h and T were recorded for this valve’s reading.
The CV’s reading was then maintained with a step change in
the HV’s reading (3/5). A steady state was achieved and the
readings of h and T were recorded. The t = 0 configuration
was then restored, that is, the readings of CV and HV were
returned back to 3 and 2, accordingly. The system was then
allowed to achieve a steady state. Then, a step change in
the CV to 4 was achieved while maintaining the HV at 2.
The steady-state readings of h and T were obtained for this
configuration. The system was again returned back to its t = 0
configuration and allowed to achieve a steady state. A step
input of the HV to 4 was achieved with the CV still at the
t = 0 reading obtained. At steady state, the values of h and T
17
16.5
16
15.5
Height (cm)
15
14.5
14
13.5
13
12.5
12
991 1191 1391 1591 1791 1991 2191
Time (s)
FIGURE 5 | Height vs. time for a step change of CW.
Shanmugam
for this configuration were recorded. Finally, the system was
returned to its initial (t = 0 configuration) state (5).
Results and discussion
The experiment dealt with controlling both the tank level and
the temperature of the water tank. The manipulated variables
were the CW valve and the HW valve. Through inserting an
input signal into the system, the output was observed as it
changed from the initial steady-state conditions to the final
steady-state conditions. Then, the system was taken back
to the previous steady-state conditions in order to simulate
the single-step input and observe the output with respect to
that input. The data collected for this lab were obtained as
a spreadsheet, and multiple figures were obtained in order to
calculate the gain of the system, the delay of the response, and
the time constant of the output.
Figure 5 represents the tank level output as the system
experiences a single input through the CW valve. The input
signal is a signal step; therefore, the U(s) term is expressed in
the following equation:
1
U(s) = (1)
s
Equation (1) represents the single input signal of the system.
The output is considered, for this case, as the water level in
the tank. Furthermore, the transfer function for this system is
considered the first order since it is a water tank. Moreover,
as observed by the curve, the output changes immediately
as the input changes. Therefore, the time delay for this step
response is zero. The time constant for the transfer function
is the time required for the system to reach 63% of its
final value. By implementing the final value theorem for this
system, the final value is equal to 4.343 cm if we assume that
the initial steady-state conditions are set to zero. The 63% of
that value is 2.736 cm with respect to the initial steady state.
By observing the curve, the time constant that corresponds
to (2.736 + 12.356) cm is equal to
τ = 1130 − 991 = 139 sec (2)
Furthermore, the final value theorem represents the MK
parameter. While noting that M is equal to 1, then the gain
of the system, K, is equal to 3.343 cm. Therefore, the transfer
function equation, G(s), is represented as follows:
K 3.343
G(s) = = (3)
τ+1 139.s + 1
Height
Poly. (Height)
Equation (3) represents the transfer function of Figure 1. If
it is compared with the MATLAB simulation, the theoretical
2391
plot, as shown in Figure 2, overlaps with the experimental
lab. This strongly demonstrates the accuracy of the transfer
function obtained for the system output.
10.54646/bije.2022.06 31

FIGURE 9 | Height vs. time for a single-step change in the HW valve.

FIGURE 6 | MATLAB simulation with respect to Figure 1.

31.5
31
30.5
30
Temperature (C)

29.5
29
28.5
28
27.5 Temp.
27 Poly. (Temp.)
26.5
26
991 1191 1391 1591 1791 1991 2191 2391
Time (s) FIGURE 10 | Temperature vs. time for a single-step change in HW.
FIGURE 7 | Temperature vs. time for step change of CW valve.

In addition, the transfer function, G(s), is represented as

31.5 follows:
31
30.5 K −2.825
30 G(s) = = (5)
τs + 1
Temperature (C)

29.5
121.s + 1
29
28.5 Figure 9 represents the tank level with respect to time.
28 Temp. A single input signal represents the U(s) of the system. The
27.5
27 Poly. (Temp.) output, Y(s), is the tank level. Furthermore, the transfer
26.5 2 per. Mov. Avg. (Temp.)
26
function, G(s), is considered the first order since the behavior
1491 1591 1691 1791 1891 1991 of the output portrays the behavior of the first-order system.
Time (s)
The behavior of the output response conveys many futures
FIGURE 8 | Temperature vs. time for step change of HW valve.
of the system. The delay, the gain, and the time constant are
figured out by recognizing multiple futures of the curve.
Figure 7 represents the change in temperature with respect Since the output immediately changes as the input
to time. The single-step input for this first-order system is the function changes, then there is no observed delay with
change of flow rate of CW. respect to the output response. Furthermore, the gain of the
As observed in Figure 3, the delay of the output response system is represented by the parameter K. Since M is equal to
is zero, because the output changed immediately once the 1 in this scenario, then K is equal to the FVT. Nevertheless, it
input changed. Furthermore, the system gain of the system is essential to nullify the steady-state conditions and set them
is represented by the FVT. The FVT, as shown in Figure 3, equal to zero. Once it is done, then the FVT is calculated to
is equal to −2.825◦ C if the initial conditions are set to zero. be 2.502 cm. Furthermore, the time constant for the output
Therefore, since the FVT is equal to MK, then K is equal signal is equal to 121 sec. Thus, the transfer function is
to −2.825◦ C. Furthermore, the time constant represents the represented as follows:
time the system takes to reach 63% of the final value. Since K 2.502
63% of the FVT equals −1.779◦ C, then the time constant is G(s) = = (6)
τs + 1 121.s + 1
as follows:
Figures 7, 8 demonstrates the relationship between the
τ = (1112 − 991) = 121 sec (4) temperatures of the tank with respect to time. The
32
manipulated variable that is changed in this case is the
HW valve. The delay of the response, the gain of the
system, and the time constant of the output function are
obtained by studying Figures 7, 8. The data for this system
are not accurate enough to obtain the required values
through visual reading. Therefore, a program called Minitab
is utilized in order to obtain the average values with respect
to specified time ranges.
The delay of the response, the gain of the system, and
the time constant of the output are calculated. The transfer
function demonstrates the future of the first-order function.
Furthermore, since the input function is as represented in
Equation (7), then the FVT is equal to K, since M is equal to 1.
M 1
U(s) = = (7)
s s
Nevertheless, it is required to nullify the initial steady-state
conditions and set them to zero. Therefore, the gain of
the system, K, is equal to 1.834◦ C. Moreover, there is no
observed delay with respect to the output response; therefore,
the delay is set to zero. Furthermore, the time constant of the
output function is equal to the time it takes to reach 63% of
the final value. Therefore, the time constant is calculated to
be 140. From these data, it is possible now to get the transfer
function, G(s), for the first-order system as observed in
Equation (8).
K 1.834
G(s) = = (8)
τs + 1 140.s + 1
Since all of the transfer functions obtained represent the
first-order systems, then it would be enough to use a
proportional, P, controller. That is because overshoots,
damping coefficients, and oscillations will not be present in
the first-order system, which diminishes the reasons behind
using either the integral, I, part or the derivative, D, part.
Shanmugam
Conclusion and recommendations
The experiment dealt with the behavior of the controlled
variables with respect to the manipulated variables of the
system. Furthermore, a mathematical model was required
to represent the transfer function of the system. The
transfer function, G(s), demonstrated the characteristics of
the system. The delays, the gains, and the time constants
of the system are calculated for each step change. The
transfer functions were obtained along with the parameters
for each curve. A comparison is done between the parameters
obtained for the different magnitudes of the input signals.
This is done in order to eliminate any discrepancies or
to determine if there was any present. It is recommended
to implement new control devices. A long time constant
of 2.33 min was observed. This resulted in longer times
in order to achieve the steady-state conditions. It is also
recommended to have the drain valve manipulated. This will
introduce new behaviors observed into the system and will
further enhance the understanding of the control concepts.
References
1. Stephanopoulos G. Chemical Process Control. (Vol. 2). New Jersey, NJ:
Prentice Hall (1984).
2. National Instruments. What Is Lab VIEW? (2013). Available online at:
http://www.ni.com/labview/ (accessed October 30, 2013).
3. Seborg DE, Edgar TF, Mellichamp EA, Doyle FJ. Process Dynamics and
Control. 3rd edn. New York, NY: John Wiley and Sons Inc (2011).
4. Harriott P. Process Control. New York, NY: Tata McGraw-Hill Education
(1984).
5. Swaffar CM. Developing Process Control Experiments for Undergraduate
Chemical Engineering Laboratories. Fayetteville, AR: University of
Arkansas (2013).
6. National Instruments. Lab View Software Manual National Instruments.
Austin, TX: National Instruments. (2013).